HNHacker News
TopNewBestAskShowJobs

brianchu

2,341 karma · joined July 15, 2012

Looking for the next thing.

Worked on machine learning for medicine at Cardiogram. Studied CS at UC Berkeley. Former undergrad researcher and teaching assistant for Berkeley's ML class. Former intern at Twitter on Ads Analytics Infrastructure.

Email: bc@brianchu.com

github.com/bchu

twitter.com/brrrianchu

submissionscomments
brianchu··on Why Are American Colleges Obsessed With 'Leadership'?
I currently attend UC Berkeley and graduated from a fairly affluent top US high school a few years ago. Meaning: everyone was very concerned (I won't say obsessed) about college admissions.

This focus on leadership in admissions is very well known, and manifests itself conspicuously in competitive high schools. There's a considerable amount of unspoken pressure to get "leadership" positions in school clubs, do "leadership" activities in community service, do varsity sports or other leadership in sports, and participate in ranked competitions. The list goes on and on. This breeds a lot of cynicism about people's motivations, especially as people try to game the system (with empty leadership positions). What starts to happen is that when people hear about a high schooler who goes off and founds a non-profit and does work in Africa (or something like that), people start to wonder whether that person did it just to improve their admissions chances. The sad part is that this person could very well be genuine.

EDIT (reply to below): Sometimes those people are college admissions officers. I'm in college now, so I don't worry much about that stuff anymore. I just recognize that the system is screwed up.

brianchu··on How Quora Onboards New Users
Despite all the really negative comments (and yes, I agree that Quora shouldn't hide its great content from guests), Quora is fairly valuable for me. I spend more time on Quora than I do on Facebook or Twitter; the only site I spend more time on is HN.

The reason is that Quora has a lot of really reputable, valuable, and domain-specific contributors, and it's easy to identify answers from people who have credibility on a subject. On HN there's a lot of noise, and there's a really common phenomenon of people bullshitting about subjects where they really have no credibility (usually the most controversial topics on HN like politics, economics, venture capital). The same is true for Quora, but on Quora you can identify people's domain-expertise better (it's posted next to each answer), and the best contributions percolate to the top much better than on HN. It's interesting to note that one thing people have always wanted on HN is the ability to follow certain users and certain topics, which is something Quora does.

For just the startup/tech world, there are tons of contributions from experienced VCs, angel investors, experienced entrepreneurs, etc. Outside of tech, you get pleasant surprises like former police officers answering questions about policing, movie directors answering questions about movies they've directed, etc.

Quora is still young, so there are a lot of quality early adopters. Going forward, it remains to be seen whether the quality will degrade as Quora gets more popular.

The biggest problem with the onboarding is that most of my value is through following "famous" (credible) people, I get very little value from following topics or questions, because when you start doing that you get lot's of noise (random or semi-credible people answering). This doesn't seem to be something emphasized in the current flow.

EDIT (reply to below): There's a lot of noise. There are a lot of gems, too, but it's easier to identify the gems. I don't think of things in terms of inside vs. outside the so-called SV bubble, so I don't really share that sentiment. An experienced entrepreneur is still credible to me, even if they are in SV.

brianchu··on How the Internet Ruined San Francisco (1999)
It's a matter of perspective. The grass is always greener on the other side, and I think it's unrealistic to pine for an idealistic childhood.

I was born and raised in Palo Alto and I go to UC Berkeley now. I spend a lot of time in SF. A lot of what you said relates to my childhood. But a lot of what happened was the result of my choices. I love technology and I love the culture here. One thing that helped me before going to college was taking a year off to work, which really helped me figure out what was important to focus on and what isn't important (not important = grades. important = meeting people). Now that I'm in college, I could easily choose to party all the time and slack off. But that's just not me. Plus, I've had fun at parties/kickbacks in SF before.

As a side note. I use "Silicon Valley" to refer to the SF Bay Area (SF + East Bay + geographic SV). To me, Silicon Valley is a term for the entire technology region, much like "LA" describes the metropolitan LA area, and "Wall Street" describes NY finance in general.

brianchu··on CoffeeScript's Time is Waning For Me
Also relevant: a big question is how CoffeeScript is going to deal with ES6 and the overlapping syntax between ES6 and CS, which may involve breaking backwards compatibility or leaving out ES6 features.

https://news.ycombinator.com/item?id=6078969

jashkenas > So in general, what I'd like to see mainline CoffeeScript do, is adopt useful ES6, ES7, and ES8 features, as soon as they land in enough platforms/browsers to be widely useful (yield may be at this point already, or very soon), and to continue to try and find a pleasing minimalist syntax for writing and reading them. If this means taking the ES* syntax wholesale, that's fine. If it means minting a new (and hopefully, but arguably, nicer) syntax, that's fine too. And if it means breaking backwards compatibility, that's also alright, as the compiled lowest-common-denominator JS output will still be perfectly compatible with the new stuff.

brianchu··on Ask HN: How would you teach web development to college students?
I would focus on JavaScript. No CoffeeScript [1]. Start with basic HTML and CSS on the frontend; nothing fancy with HTML/CSS yet, just a couple of workshops. Then basics of JavaScript. Then jQuery (leave the DOM for another day). Then you can consider branching out into select (can't cover them all) advanced topics like D3.js, HTML5 canvas, SVG, WebGL (using three.js).

You may find that that's all you can cover in a semester. If not, I think it's important to also cover the server-side. Start with jQuery AJAX calls against an already created RESTful API to teach the basic request-response HTTP model. Then have students implement the API (no database needed) in Node.js. If you still have time (I doubt it), introduce the concepts of databases with MongoDB [2].

Python and Ruby are great ecosystems, but if you still have to teach JS, I feel like it's a bit hard on beginners to juggle multiple languages. Also, I'm divided on the utility of teaching a huge framework like RoR/Django vs. a lighter framework like Sinatra, Flask, or Express.js (node).

[1] You want to minimize cognitive overhead, especially with 2/3 non-CS-majors. And since anyone learning CoffeeScript needs to know JS, you might as well start with JS.

[2] MongoDB is a lot quicker for hacks than SQL.

brianchu··on Ask HN: What new skill or tech you intend to master in 2014?
I'm a student at Berkeley. Here's my list:

learn Vim.

master iOS and release several apps.

learn devops / sysadmin.

gain experience with SQL/Postgres (as opposed to just hacking together something with MongoDB, the easy way to do things)

My background is in iOS and JavaScript web (Angular.js, Node.js).

brianchu··on Ask HN: I just dropped out of college, should I do a coding bootcamp?
1. Java and C++ are quite foundational, I think. I don't think it is a problem at all that you don't learn version control in school. Version control is quite simple conceptually and is a simple matter of memorizing the API/commands.

2. I'm familiar with the programs that you mentioned and the general hiring climate in SV/SF. Email me (listed in my profile) for a deeper discussion on your options.

brianchu··on Ask HN: I just dropped out of college, should I do a coding bootcamp?
One data point: Mark Zuckerberg did not drop out of Harvard when he left to do Facebook. He took a leave of absence and he did not permanently drop out until November 2005, by which time Facebook was well on its way to success, with a 100MM valuation and a position as the 10th most trafficked website.

So at least for him, he was probably pretty sure of his success (a cushy acquisition as the very least).

http://www.thecrimson.com/article/2005/11/1/zuckerberg-to-le...

brianchu··on Ask HN: What sites do you contribute to the most, besides HN/Twitter/FB?
I contribute a bit to Quora. Despite the bad rap that Quora has on HN, Quora does a really good job of curating posts, and there are a lot of extremely high quality posts related to software and startups. It's a lot less nitpicky than HN because anecdotes and opinions are perfectly fine on Quora.
brianchu··on Ask HN: Guidelines for building a developer's personal website
1. Have a blog.

2. Have an About/Bio page.

3. Have a projects page that showcases your projects, ideally with pictures.

4. Links to Github, Twitter, LinkedIn.

5. Responsive (lots of mobile visitors).

Bonus: Make sure everything is SEO-ified and social-media-ready (e.g. use Google authorship, make sure the right meta tags are present for SEO and social media sharing). Overkill-bonus: Add a contact form to reduce the friction of people reaching out to you. You can see my website, http://www.brianchu.com, for an example.

I'm a university student, yet engineers at tech companies have in fact reached out to me through my website (contact form) before.

brianchu··on Ask HN: What habit have you dropped or picked up that improved your life?
There's a big difference between reading the news every day, reading the news every week, and reading the news every month.

I used the read the news every day (and reading news magazines weekly); what ended up happening is that I also ended up reading through a lot of fluff and reading about a lot of inconsequential mere-fact-stating news. Even if you read the news every week, there's a ton of repeated analysis that you have to filter out. I ended up realizing that this was a legitimate waste of time.

Now I just read the news maybe every month. I'll usually only read long-form synthesis/analysis articles, rather than the kind of stuff that comes in through AP/Reuters. The exception being news articles coming in through HN, of course.

Going back to your analogy, while I'll keep myself updated on the major movements in the civil war in Syria, what I don't need to read about is how 12 people died in a bomb blast on Tuesday in <city> or something extremely-low-level like that.

brianchu··on Show HN: My first iPhone game, Devils in Heaven
I've not used both, but keep in mind that SpriteKit is only for iOS 7.

Cocos2d-iphone is also really mature, compared to SpriteKit.

If you want something cross-platform, look into cocos2d-x, which uses C++, but you can use their JavaScript bindings.

This is all for 2D games. For 3D look into Unity.

brianchu··on Show HN: My first iPhone game, Devils in Heaven
Yep!
brianchu··on Show HN: My first iPhone game, Devils in Heaven
Yep, skywing and quantumpotato have the general gist of it.

When I first started I thought it was going to be really simple, but it turns out that building a game can be quite tough. Compared to building a web or mobile app where you have lots of generic scaffolding available to you (web frameworks like RoR, stuff like UIKit for iOS that provide all the UI elements for basic apps), for a game your code has to manage every image and every interaction, so all the special cases and checks and loops add up pretty quickly.

Plus, implementing stuff like ragdolls and physics adds a lot of complexity.

Objective-C is pretty verbose, so that adds to the total.

brianchu··on Show HN: My first iPhone game, Devils in Heaven
cocos2d-iphone is really simple and easy to use; it's my own code that messes things up :).

The thing that added the most complexity was Box2D. I'd probably use Chipmunk if I were to write an iOS game again, simply to keep everything in pure Objective-C (due to available Obj-C bindings).

brianchu··on Jam vs ender vs colo vs component vs browserify vs bower?
Note that bower is extremely lightweight; all it basically does is pull repos from github. So it can hardly be considered a package manager. The rest all offer various features above and beyond bower, AFAIK.

The reason bower is so popular is because it doesn't do very much.

brianchu··on Startup CEOs: you are overestimating your tech talent
Some of the technical challenges I can think of:

1. Version control.

2. LAN sync.

3. Dropbox probably also does a lot to try and minimize their S3 costs - so probably a good deal of compression/duplication-detection on the backend.

4. Any and all networking issues. Even if using OSS gets you 98% of the way, Dropbox probably gets you 100% of the way.

5. Performance. I don't know what Dropbox is doing, but about a year ago I tested out Google Drive, SkyDrive, Dropbox, and SugarSync. I didn't want to use Dropbox because it gives you the least free storage, but Dropbox was consistently 2x faster than the competition (without using LAN sync). That makes a huge difference so I switched to it, and if Dropbox engineers are beating out Google and Microsoft engineers, they're clearly doing something special.

brianchu··on Snapchat is Intrinsically Worthless
Since a lot of people have trouble imagining why Snapchat is popular, I wrote a blog post on it a while ago: http://www.brianchu.com/blog/2013/09/11/photo-embarrassment/.

One thing that bears noting is that Snapchat is closer to mobile messaging apps like WhatsApp than it is to photo sharing / social networking apps like Instagram or Facebook.

brianchu··on Ask HN: Why does DNA have only 4 nucleotides?
Several thoughts (from someone with only intro-level biology, fwiw):

1. I think a lot of biological why's might be fundamentally un-answerable except for "pure statistical randomness." That's just the branch that evolution happened to fall towards.

2. Or perhaps it is answerable. In terms of costs and benefits, it could be that the cost of the additional complexity in replication, transcription, translation and all the tasks DNA serves outweighs the error-reduction that is possible with additional base pairs (either by allowing triplets/quadruplets/etc or by allowing greater redundancy in mapping to proteins). There are probably diminishing returns in error reduction after having two base pairs.

brianchu··on Snapchat is a fad, not a business. Do not glorify.
A lot of people don't seem to get Snapchat, so I wrote a blog post about it here: http://www.brianchu.com/blog/2013/09/11/photo-embarrassment/.

Have people actually met/seen others using Snapchat? I'm in college so tons of my friends use Snapchat regularly (about every couple of hours). Snapchat is essentially used for picture messaging of frivolous images, almost like a picture-status-update of something funny or current. Nobody I know really uses it for "dirty" pics, though unflattering pics (of yourself or others) are fairly common. It's almost a foil to Instagram - Instagram encourages you to post by making your photos permanent but beautiful; Snapchat encourages you to post by making your funny/ridiculous/unflattering photos ephemeral. See the difference? It's ephemeral vs permanent, funny/ridiculous vs beautiful.

brianchu··on Pinterest Raises $225 Million at $3.8 Billion Valuation
> Metrics only indicate what people think they want, not what they need. Metrics are only accurate if you look for trends. Long-lasting businesses don't rely on trends.

This really does not make any sense. Metrics as a general concept can and do encapsulate all the positive things you just mentioned (long-lasting value, what people need, etc). Just because a subset is shitty/misleading does not invalidate my argument. Obviously an investor must sift shitty metrics from good metrics, and shitty teams (that appear good) from good teams.

> Facebook, just like AirBnb, make a lot of sense. You don't need metrics to understand that.

Firstly, hindsight is 20/20, so that's easy for you to say. Obviously everyone now believes Facebook is a good idea; the same is not true back in 2004. Good investors recognize that they cannot necessarily tell the good ideas from the bad ideas, so they focus on team/metrics. The most revolutionary ideas typically seem like bad ideas.

> Do you want to live in a world where investors all throw their money at the next Instagram? Or do you want to live in a world where they invest in the next Xerox Parc?

You have not provided a reason we cannot have a world with both. Those are not mutually exclusive.

brianchu··on Pinterest Raises $225 Million at $3.8 Billion Valuation
At several Startup School talks, Ron Conway, probably the most successful angel investor in SV, has frequently said "you never argue with metrics." In a talk at Startup School 2012, he said that he "didn't get" Facebook, but he invested (I think it was the round after Peter Thiel's investment, so very early) after seeing the metrics, "way up and to the right."

Somewhat similarly, USV has a well-known blog post detailing how they really liked the Airbnb team, but they "didn't get" the idea, missing out from investing at the seed stage in a multi-billion dollar company.

There is a reason that metrics (growth) and/or a stellar team are commonly cited by investors as being more important than (getting) the idea.

Notwithstanding that Pinterest is a great idea.

brianchu··on Turn mailto links into clean contact forms
Here's a roll-your-own alternative: use an embedded Google form and style it to fit your website's theme. That's what I do at http://www.brianchu.com (Contact Me button at the left). Works like a charm and I get an email notification every time someone submits the form.

A contact form is superior to publicly displaying an email address because 1) if you think about it as a conversion funnel, that's one less step for the person to reach out to you, and 2) a contact form makes it really clear that you're open to being contacted; some people are hesitant to email you out of the blue.

brianchu··on AngelList raises $24 million at a heavenly valuation
The interesting thing is that, from when I talked to the guys at AngelList, the bigger market (with larger monetization potential) is not its funding platform. It's the hiring platform.
brianchu··on Why Aren't Schools Grouping Kids By Intelligence Level?
Students in public schools are routinely ranked by intelligence. There's the entire concept of "class rank," in which students are in fact put in ranks by their GPA (a proxy for intelligence) and from which a valedictorian is chosen. Then there are standardized tests, which are fairly analogous to IQ tests.

I think it's a rather serious overreach/stretch to link this to the broader workings of democracy.

brianchu··on Ask HN: Can someone please write a Chrome Extension to kill new Gmail Compose?
On it.
brianchu··on Ask HN: 1% equity vested over 4 years, is it okay for first employee?
1. Agreed with tptacek. 1% is within the "standard" range of equity from what I've heard in SV. But that range is highly variable: 1-5%, but even 0.1-10% might be justified. But we have nowhere near enough information to judge whether that kind of an offer is a good offer for your specific, individualized case. Some questions to consider: What kind of salary are you making? What salary could you make at a BigCorp? When was their last round of funding, how much did they raise, who did they raise from, and what was the valuation? What kind of shares were issues to the investors (preferred, participating preferred, bla bla bla etc.)

2. Approach this like an investor. You are "investing" a below-market salary (and, if you are going to be working longer hours than at a BigCorp, you have to factor that into your calculations), and in return you are getting equity. The easiest way to calculate the dollar value of your equity is valuation * equity (then divide by four since this is vested over four years), and that is the rough cash equivalent of your equity). But depending on what kind of stock your investors got (preferred stock vs. the common stock or stock options you get), you might want to discount this valuation by 10-20%.

3. Valuing startup equity is a huge topic, there are entire HN threads dedicated to this: search HN for "startup equity." Here are some good articles: http://mba-mondays.pandamian.com/tableofcontents/ (read the Employee Equity sections), pg's essay on this (http://www.paulgraham.com/equity.html), an article on someone's negotiation process (https://keen.io/blog/29904565692/how-i-negotiated-my-startup...), and another summary article (http://www.payne.org/index.php/Startup_Equity_For_Employees).

brianchu··on Ask HN: How would you defend Python against Node.js in a SaaS web app?
If you have a sufficiently JavaScript-heavy client-side codebase, there is a very strong argument for essentially unifying the server-side and client-side tooling. Being able to write and run unit tests in exactly the same way (jasmine or mocha), being able to use the same IDE or text editor plugins, being able to use the same package manager (npm), being able to move client-side and server-side developers relatively effortlessly, being able to use the exact same build tools (grunt or something else). You also get to share code between the client and server (this isn't a huge benefit, though).

As a bonus you get the fact that node.js can scale better, though this should not be a huge concern.

Python has a real advantage if you're doing lot's of data analysis / machine learning, as the node.js ecosystem for those applications is nearly non-existent.

brianchu··on The HN Digest trial
It isn't the article that I usually spend the most time reading. It's in the comments where most of HN's learning value lies, and thus where most of my time is spent. If the summaries included an outline of the opinions/takeaways/great resources mentioned in the comments, that would make the product that much more worthwhile.

I also tend to hoard links. If there's a great thread on Haskell, for example, I'll bookmark it (save it to Pocket), with the intention of going back to the thread if/when in the future I decide to learn Haskell.

Ultimately, though, the problem is that whether or not I pay for this depends on whether this tool actually weans me off HN. If the summaries are so high quality as to get me off HN (except to post comments myself), I would pay $50 a month for this. If I still find myself reading threads by myself on HN, I wouldn't pay anything.

brianchu··on The Real Value of Stock Options
1. I really disagree with the assumption that you're not going to get more than 0.5% at a seed-funded startup. It is really common, at least in Silicon Valley, for an engineer at a seed-funded startup to receive more than 0.5%, especially if they are taking a below-market salary. In fact, if you are one of the very first employees, you should definitely be getting more than 0.5%.

2. It is also very common, and by no means something particularly extraordinary, for an early employee (post-seed) to receive stock grants instead of stock options, which makes the situation even more favorable (especially if you declare the grant as income to the IRS early).

← PreviousPage 5 of 10Next →